How does Motivational Interviewing (Ml) work? Although studies consistently show positive effects of Ml, research has found few relationships between client characteristics and outcomes. Possible mechanisms of action (MOA) include Ml therapist behaviors, client behaviors, and interpersonal interactions. The investigators completed a randomized controlled trial of the efficacy of an adaptation of Ml, Project CHOICES. Project CHOICES tested a four session, manual-guided intervention targeting risk for alcohol- exposed pregnancy (AEP). There was a small to moderate effect of the intervention on risk drinking and inadequate contraception and a moderate effect on AEP risk among 415 risk-drinking women drawn from 2 high risk samples in each of 3 cities. The proposed study will use Project CHOICES'audiotapes of therapy sessions and its outcomes dataset. We will rate audiotapes to capture therapist and client behaviors and interpersonal interactions, and merge this process data with the existing dataset. We will then test a number of hypothesized MOAs of Ml. Specifically, we will examine therapist behaviors, client behaviors, and nterpersonal interactions, evaluate their inter-relationships, and their relationships to outcomes. We will test the Ml behaviors of CHOICES intervention therapists using the Motivational Interviewing Treatment Integrity (MITI) code. We will test client behaviors (change and resistance talk) using the Motivational Interviewing Skill Code and the Client Resistance Code. We will examine interpersonal interactions on an act-by-act and global basis using validated measures to determine the relationship of interpersonal interaction variables to therapist behaviors, client behaviors, and outcomes. Using interpersonal theory to examine Ml is an nnovation that will allow us to depict Ml therapist behaviors and client behaviors in interpersonal circumplex space. Lastly, we will develop an explanatory model of the MOAs of Ml by evaluating competing mediator and moderator models of the impact of therapist Ml behaviors, client behaviors, and interpersonal nteractions on each other and on final outcomes. The public health impact of this project will be to specify the necessary and sufficient components of Ml that will allow science-based improvements to clinical practice. Once these components are identified, new, science-based interventions could be developed for a range of health-threatening behaviors for people with chronic illnesses including addictive disorders.